REAPPORTIONMENT AND PARTY REALIGNMENT IN THE AMERICAN STATES 1

REAPPORTIONMENT AND PARTY REALIGNMENT IN THE AMERICAN STATES 1 Stephen Ansolabehere Department of Political Science Massachusetts Institute of Technology James M. Snyder, Jr. Departments of Political Science and Economics Massachusetts Institute of Technology August, 2003 1 Professors Ansolabehere and Snyder gratefully acknowledge the support of the National Science Foundation (SBER-9631640) and the Carnegie Corporation. We wish to thank Philip Burrowes for his research assistance. Paper presented at the Annual Meeting of the American Political Science Association, Philadelphia, PA, August 30, 2003.

Abstract Malapportionment of state legislatures before the mid-1960s gave urban and surburban voters much less representation than they deserved. This paper documents that suburban and urban voters had markedly different policy preferences, party identifications, and partisan voting behavior than rural areas, which were overrepresented. However, the patterns are not uniform. In the Northeast and North Central, the suburban and urban under represented areas were much more Democratic than rural areas. In the South and the West, the rural areas leaned more Democratic than the urban and suburban voters. Policy preferences split differently in the Northeast and North Central than they did in the South and West. Urban and Suburban voters were much more liberal on social welfare and economic policy than rural voters in those areas. In the South and West, few differences existed across locales. On only one issue did the urban and suburban areas have more liberal attitudes throughout the nation: racial politics. Court-ordered reapportionment, thus, increased the political weight of liberals and Democrats in the Northeast and North Central, but not in the South and West. Consistent with Erickson (1973), reapportionment moved the median voter in all regions to the left on issues of civil rights and racial policy.

1. Introduction Reapportionment of state legislatures during the 1960s radically altered representation in the United States. Throughout the first half of the Twentieth Century, most state legislatures either required representation of area as well as people or neglected to draw new district boundaries, despite state constitutional requirements for population based representation. As a result representation in state legislatures failed to reflect much of the growth in urban and suburban areas that occurred during the late 19th and early 20th centuries. By 1960, dramatic differences in existed in at least one chamber of almost all state legislatures. In California, for example, Los Angeles county had one state senate seat for its 6 million people and the three smallest counties in the state, with combined population of 17,000 people, shared a senator. In Connecticut, the City of Hartford had two state representatives for its 180,000 residents, while Hebron also had two representatives for its 1,800 residents. Through a series of signficant court cases, beginning with Baker v. Carr, the U.S. Supreme Court forced the states to eliminated these disparaties by the end of the 1960s. 1 The sudden decline in rural political power in state legislatures had broad effects on public policies. Equalization of representation altered the distribution of public spending across areas within the states. Over-represented areas had long gained a disproportionate share of public expenditures because of their advantaged political positions. That vanished once representation was equalized (Ansolabehere, Gerber, and Snyder 2002). It was natural to believe that the liberal urban agenda would succeed in other policy matters as well. Surprisingly, a broad shift in public policy in the states cannot be traced to reapportionment. Dye (1965), Frederickson and Cho (1970), and Jacob (1964), however, find no or little evidence that malapportionment affected the overall liberalness of state policy, including overall levels of expenditure and labor regulation. The exception is civil rights legislation. Erikson (1973) documents increased support for civil rights in state legislatures as a result of reapportionment. In this paper, we examine the effects of reapportionment on the political parties. At the time, it was conjectured that Democrats and liberals would see the greatest political gains, because urban areas tend to be the most Democratic and usually had the least state legislative representation. Democratic, labor and liberal political organizations provided much of the political activism in 1 Lowenstein (1995, pages 71-113) provides an excellent summary of the cases, their progression, and the legal and constitutional issues involved. 2

support of reapportionment. Erickson (1971) examines the effects of malapportionment on party control of non-southern state legislatures that were substantially malapportioned. In general, he concludes, Democrats tended to gain; however, only half of the chambers analyzed showed substantively large effects and some states saw significant Republican gains. (See also Roebeck 1972.) Subsequent studies have found similarly small net gains of the Democrats in the wake of reapportionment. Across the nation, Democrats seemed to have gained about 10 percent more state legislative seats (Egan, Kousser, and Persily 2002). The lack of party effects has been a cause of some debate, with partisan gerrymandering often blamed for the weak Democratic gains (Cox and Katz 2001; Egan, Kousser and Persily 2002). Two puzzles emerge from past research. First, why did the Democrats make only modest gains following Baker, given the enormous under representation of cities? Second, why are the policy effects of reapportionment limited to civil rights and the distribution of public expenditures? Uneven policy changes and weak Democratic gains reflected, we believe, the nature of malapportionment throughout the country prior to Baker v. Carr. While partisan gerrymandering and related monkey business probably contributed some, much of the pattern of policy shift and partisan shift can be understood in terms of three factors: who was underrepresented, where, and what did they believe. To answer these questions we examine the contours of electoral behavior and of citizens policy attitudes across the regions and parties in the decades leading up to the implementation of the one-person, one-vote standard. We examine aggregate data on state election returns and the structure of legislative districts to measure the partisan effects of malapportionment. We examine the National Election Surveys from 1952 to 1968 to map the policy preferences of urban, suburban, and rural voters living in different regions. There was not one realignment revolution, but many. Malapportionment in the state legislatures regularly followed the contours of population, with rural areas having disproportionately more state legislative representation. However, the partisanship and political orientations of rural, suburban, and urban communities varies across states and regions. We discern four distinct regional patterns of partisan under-representation that are attributable to malapportionment. In the South, malapporitonment advantaged the Democrats, because rural areas voted much more Democratic than the cities. In the Northeast and North Central, malapportionment tended to advantage Republicans, because rural areas in these regions voted heavily 3

Republican while the cities voted Democratic. In the West, a more mixed picture emerges, and the differences between urban and rural are less pronounced than in other regions. Political orientations and policy preferences also varied across regions and locales. Southern rural voters, who were over represented throughout the South, tended to be very conservative; Northern urban voters, who were under represented in their regions, tended to be very liberal. The differences between these two sorts of voters are well known, but not exactly relevant. The more meaningful comparison is within region. How did these voters compare to other partisans across geographic locales within their respective regions? Were urban and suburban Southerners, for example, more liberal thant rural Southerners? Such would have to be the case for reapportionment to affect public policy by realigning the electorate represented in the state legislatures. Some important differences did exist; however, the patterns are such that the policy implications of reapportionment varied across regions and across areas of public policy. Both within the parties and in the electorate as a whole different political geography correlated differently with ideological belief and policy liberalism across regions. In the Northeast and North Central, reapportionment had the greatest potential to shift policy to the left; in the West, there was no such potential. The potential to shift policy, we document, came from two engines. Fist, reapportionment had the potential to shift the median voter in the state legislative electorate as a whole. The average voter in many regions was much more liberal than the over represented rural voter. Second, reapportionment had the potential to move the political parties. Urban and rural voters within the Democratic party differed substantially on most issues of the day. The Republicans were not similarly split. Reapportionment in the mid-1960s likely fueled the divisions within the Democratic party divisions over race, labor relations, education and economic policy that events and organizations were pushing to the fore of the national political agenda. In the pages that follow we document these patterns using a mix of aggregate and survey data. Our goal is less to estimate the effects on specific policy changes and more to document the patterns of malapportionment as they relate to the representation of political preferences. Ultimately, we argue that reapportionment produced four different regional patterns of partisan realignment. In the Northeast and North Central, reapportionment shifted politics toward Democrats and the left. In the South, reapportionment shifted politics toward the Republicans, but not assuredly to 4

the right and on issues of race the shift was in the liberal direction. In the West, reapportionment had little immediate partisan and ideological impact. 2. Representation and Partisanship How did inequalities of representation relate to partisanship? Malapportionment produced partisan advantages to the extent that rural and urban areas within states and regions had differing party attachments. By far the most important factor explaining malapportionment was population. Rapid urban population growth created a rural backlash in the early 20th Century that produced constitutional and legal measures designed to guarantee over representation of rural interests. Typically states adopted rules that gave each county at least one seat in each chamber. Malapportionment, then, advantaged partisans in rural areas, whatever their political leaning. (See McKay 19xx, Dixon 19xx, and Ansolabehere, Snyder, and Woon 1999.) State legislatures and constitutional conventions sometimes magnified these advantages further through gerrymandering that gave the minority party especially few seats. New York state is a case in point. In 1894, New York convened a constitutional convention to revise its constitution. The convention was dominated by Republicans and wrote into the new constitution boundaries for the new legislative districts. A coalition of outstate Republicans from rural and urban areas (especially Buffalo) and rural Democrats allotted New York City many fewer seats that it deserved. Republicans from New York City went along with the plan, which was approved in a straight party vote, because the plan strengthened the party in the state legislature. The plan also Gerrymandered the city to create some Republican seats, one of which was seat four city blocks wide and 80 long. 2 One measure of the relationship between partisanship and malapportionment is the correlation between relative representation and Democraticness of counties within states. David and Eisenberg (1961) construct a measure of the representation as a county s fraction of state legislative seats divided by that county s fraction of state population. If a county s share of legislative seats equals its share of the state s population then the index equals 1. Ratios higher than 1 mean that the county has more representation than it deserves; ratios less than 1 mean that the county has less representation than it deserves. We call this measure the Relative Representation Index, or 2 See Procedings of the New York State Constitutional Convention of 1894. 5

1960, immediately before Baker. 3 [Figure 1 here] RRI. The Relative Representation Index is highly skewed. To reduce the skew we convert this index to the logarithmic scale, which implies that fair representation has logarithm of RRI of 0. The Logarithm of RRI has a mean of.31 and variance of.58 for state representative elections, and a mean of.29 and a variance of.58 for state senate elections. We construct a similar measure of the propensity of a county to vote Democratic. The average Democratic share of the two-party vote in the county over the past 2 elections for President, US Senate, and governor. We calculate the Relative Democratic Vote of the county as the average Democratic vote in the county divided by the average Democratic vote in the state. Relative Democratic Vote equals 1 when a county has vote share equal to the state average. Values above 1 mean that the county is more Democratic than the state; values below 1 mean that the county is more Republican than the state. This measure is centered at 1, a variance of.02, and a symmetric distribution. For shorthand, we call this measure RDV. The correlation between Relative Representation and Relative Democratic Vote capture the extent to which malapportionment favored Democratic counties or Republican counties. We estimated the correlation between Log of RRI and RDV for the upper and lower chambers of each legislature. A handful of states, such as Connecticut, Delaware, and Rhode Island, have a small number of counties and have town level representation. We omit these from the analysis, though town level analysis yields a similar pattern. Figure 1 presents the correlations of representation and partisanship for the upper and lower chambers of each state s legislature in The correlation between representation and partisanship captures the partisan advantage created by malapportionment. Two patterns emerge. First, there is a strong positive relationship between the partisan advantage in one chamber and the partisan advantage in the other chamber. The more malapportionment advantages a party in the lower chamber, the more it advantages that party in the upper chamber. The only clear exception is California, where the state senate represented counties and the assembly represented population. The Democratic cities, especially Los Angeles, San Francisco and Oakland, were badly under represented in the Senate, creating 3 Alaska and Hawaii are omitted from the data as they became states in 1958. Nebraska is omitted because it has a unicameral legislature. 6

a Republican bias in the Senate. In the Assembly, San Francisco was overrepresented, creating a Democratic bias in 1960. Second, the regions of the country show several distinctive patterns. The South exhibits strong positive correlations between Democratic vote and representation. Malapportionment advantaged Democratic counties in the South. In the Northeast and much of the Midwest, malapportionment tended to favor more Republican counties. The West and Upper Midwest (i.e., Minnesota, Illinois, Michigan, Indiana, and Wisconsin) show weaker partisan effects overall. Survey data provide further evidence of the regional flavor of the partisan nature of malapportionment. The National Election Survey measures the party identification and geographic location of respondents as far back as 1952. Owing to small sample sizes it is impossible to estimate the party identifications of people in different areas within the 50 states. However, pooling the data from 1952 to 1968 we can construct reasonably precises estimates of the partisanship of urban, suburban, and rural voters in each of the four census regions the Northeast, the North Central, the South, and the West. 4 To make the NES party identification measure comparable over time detrended and standardized the traditional 7-point measure. Our measure has a mean of 0 and a standard deviation of 100. A partisanship score of 50, for example, means that an the partisanship of an area is one-half of one standard deviation above the overall mean of the item (in a given year throughout the nation). The standardized party identification measure for each type of locality and each region are shown in Table 1. Within each region party identifications differed significantly across locales. [Table 1 here] As with the aggregate data, the survey data show sharply different partisan alignments across locales in the different regions at the time of reapportionment. The Northeast and the North Central exhibited similar partisan divisions. Rural residents of Northeast and North Central, who tended to be overrepresented within their states, had very strong Republican attachments. Indeed, these were the most Republican areas in the nation. Northern suburbanites tended to identify with the Republicans, as well. Urban residents in the North, who 4 The Northeast contains CT, ME, MA, NH, NJ, NY, PA, RI, VT; the North Central contains IL, IN, IA, KS, MI, MN, MO, NE, ND, OH, SD, WI; the South contains AL, AR, DE, D.C., FL, GA, KY, LA, MD, MS, NC, OK, SC, TN, TX, VA, WV; and the West contains AK, AZ, CA, CO, HI, ID, MT, NV, NM, OR, UT, WA, WY. 7

were underrepresented, leaned Democratic. As with the aggregate data, survey responses from Northern states show a negative correlation between Democratic identification and representation (locale). Southerners had the strongest allegiance to the Democratic party relative to the rest of the nation at the time of reapportionment. The most Republican areas within the South the suburban areas were underrepresented. Reapportionment in the South would have shifted the states in the Republican direction, but even the underrepresented areas leaned Democratic. Westerners showed the least partisan division across locales. In the West Rural voters were most Democratic and Suburban voters most Republican. Reapportionment in the West, then, likely increased Republican reprsentation somewhat. Of note, we are not arguing that the under representation of one party or another through malapportionment was intentional. The correlation between party and geography and its relationship to representation was at times intended, but more often driven by other factors, or simply an historical accident. The apportionment of the Connecticut legislature dated to that states 1818 constitution before the industrial revolutin in that state. Town populations were roughly equal, and town representation saved the state the complications of a census (see ACIR Report xxxx). The partisan and ideological implications of such arrangements a century later could not be foreseen. In many states, however, malapportionment intentionally squelched the voice of some groups. The most infamous cases came in the South, where black belt counties were given less representation so as to limit the political influence and rights of black voters (Key 1949, especially Chapter 25). Black voters also tended to be Republican. In the Northeastern and Midwestern states, like New York and Minnesota, malapportionment was intended to limit the vote of urban Democrats and liberals. Finally, many states acted not against one party or another but against the growing influence of the major cities. In 1926, California voters approved representation of counties in the state senate in order to contain the growing political influence of Los Angeles, which accounted for 40 percent of the state s population by 1930. Los Angeles happened to be close to the median voter in the state in terms of partisanship in the 1920s (Ansolabehere, Snyder, and Woon 1999). Baker and subsequent cases concerned representation of population, and not the partisan effects of malapportionment. Both the survey and aggregate data reveal the varied implications for reapportionment across the country. Democrats stood to gain in the Northeast and North 8

Central but lose legislative representation in the South. It is little wonder that the Democratic party establishment in Tennessee, Florida, and other southern states fought attempts to reapportion their states throughout the 1950s and 1960s (Havard and Beth, 1954; Graham 1971). Only in the West did malapportionment have little regular relationship to representation of the parties. 3. Divisions Within The Parties Reapportionment divided the parties as much as it improved their electoral position vis-a-vis their opponents. Tennessee provides a telling pattern. In 1962, when the Court decided the case, Tennessee had not reapportioned since 1901. Population growth in Memphis, Nashville, Knoxville and Chattanooga meant that these cities had fewer legislative seats than they deserved. In addition, the rural eastern parts of the state, which had strong Republican allegiance since the Civial War, were also under represented. The central and western rural counties held the majority of seats, though they did not have a majority of the population. These counties voted overwhelmingly Democratic, and they dominated the state legislature. While Republican areas picked up some seats from these areas, the Democratic cities Nashville and Memphis gained even more. By the 1970s, the cities had become much more numerous within the Democratic party in Tennessee. This pattern varied across regions. In the South, the Border states and Mountain West, rural Democrats dominated often at the expense of urban Democrats: Tennessee is typical of the stories there. Northeastern and Midwestern states had the obverse pattern: Rural Republicans were over represented before the reapportionment revolution. Connecticut, for example, proscribed in its 1818 constitution that each town have at least one representative and no town more than two. The small towns of Connecticut tended to vote heavily Republican and their votes counted for much more than either the Democratic bastions of Hartford and New Haven, or the wealthy Republican suburbs of New York City, especially the towns of Stamford and Greenwich. How did the changing geographic composition of the state legislatures alter the parties? We examined survey responses to the National Election Survey to a range of questions concerning ideology and public policy from 1952 to 1968. The issues included Social Welfare, Government Guaranteed Jobs, Labor Unions, Health Care, Aid to Schools, Regulation of the Economy, Segregation, Civil Rights, and Religion. As with the party identification measures, we 9

standardized each measure, i.e., we subtracted the mean and divided by the standard error. The parties showed markedly different patterns. Consider, first, the Democrats. Table 2 presents the attitudes expressed by self-identified Democrats within each of the four regions and across localities on a range of domestic policy issues as well as general ideology. We denote in bold any questions that differ significantly across localities within a region. The first row of Table 2 bears a familiar pattern. Northeastern and North Central urban Democrats were the left wing of their party in the 1950s and 1960s, and Southern Democrats were the right wing. Indeed, rural southern Democrats were the most conservative group within their party. In the Northeast and North Central, reapportionment meant increasing the representation of liberals. However, in the West and South, there is no statistically significant difference on general ideology within the parties. Looking at specific questions of domestic policy reveals a deep split between Urban and Rural Democrats throughout the country. On a range of economic and social welfare policies, Urban Democrats were substantially more liberal than there rural and suburban counterparts within their own party. These differences were most pronounced in the Northeast and North Central. To the extent that the Democratic party reflected the preferences of its voters, reapportionment likely moved the Democrats to the left on social welfare and other domestic, especially in the Northeast and North Central. In the West, Urban and Rural Democrats showed no significant divisions or differences on social and economic policies. The great divide within the Democratic party came over racial politics school integration, desegregation, and civil rights. In all regions, Urban Democrats were substantially more liberal than Rural and Suburban Democrats. The division was deepest in the South, where the difference between Urban and Rural Democrats on support for civil rights legislation, school integration, and general desegregation was greatest. Battles over these issues arose long-before Baker, but, at least within the Democratic party, reapportionment shifted political weight in the liberal direction at the time that implementation of integrationist policies was truly taking hold. (See Rosenberg xxxx on the timing of integration.) In stark contrast to the Democrats, urban, sburban, and rural Republicans within each of the regions divided over few issues. Table 3 parallels Table 2, but the subset of respondents consists of Republican party identifiers. Very few issues produced statistically significant differences 10

between the over represented rural areas and the under represented suburban and urban areas among Republican voters. In the West and North Central, only four out of 38 questions showed statistically significant differences across locales. To the extent that geography correlated with policy preference, those differences appeared in the South and Northeast. Like their Democratic counterparts, urban and suburban Republicans in the Northeast were, on the whole more liberal than their rural Republicans within the Northeast. Urban Republicans in the Northeast gave relatively liberal answers on questions about health care, school aid, government guaranteed jobs, school integration, and religion. Indeed, these Republicans offered policy opinions that were more liberal than the nation as a whole, though they considered themselves to be conservative. Malapportionment effectively reduced the weight of these voters within the Republican party in the North, especially in states like Connecticut, New Jersey, New York, Pennsylvania, and Rhode Island, where there was substantial urban Republican strength. Southern Republicans, though a rare breed, showed the greatest divisions among the four regions for the GOP. Republicans were most numerous in Florida, Tennessee, and Virginia during the 1950s and 1960s. The issues that separated urban and rural Republicans in the South were not racial matters, but questions about domestic social and economic policy. And the division runs counter to the more common pattern in which urban voters are more liberal. Urban and suburban Republicans in the South expressed much more conservative attitudes than their rural counterparts on health care, school aid, government guaranteed jobs, economic regulation, and the power of the federal government. In their general ideological orientation urban Republicans in the South had the most conservative identifications of all groups. In some Southern states, reapportionment immediately benefitted urban Republicans. Again, consider Tennessee. Shelby county held at large elections for its eight lower house seats. Democrats nearly always won those seats in the 1950s, though not without clear Republican opposition. The districts created by the Democratic state legislature following reapportionment returned equal numbers of Republicans and Democrats from Shelby county and Memphis. Comparing the two major parties, it is evident that malapportionment affected the composition of the two major parties differently. Underrepresentation of urban areas lessened the political weight of urban, more liberal Democrats in all regions of the country, except perhaps the West. Locale had less clear relationship to the policy preferences of Republican identifiers. There is almost no association between geography and ideology among western and midwestern Republicans. 11

In the South, urban Republicans, who tended to be underrepresented, were more conservative than rural Republicans, and in the Northeast urban Republicans tended to be somewhat more liberal than rural and suburban Republicans. 4. Ideological Divisions The internal party divisions provide part of the answer to one of the puzzles with which we began: why was there little policy change? The parties are important in organizing legislatures. Because the parties were affected differently in the different regions, any ideological shift was at best uneven. How did these internal party divisions net out in the electorate as a whole? Table 4 parallels Tables 2 and 3 but presents the data for all respondents. Urban voters in the Northeast and North Central were most liberal within those regions. In terms of their overall ideological identifications, the typical urban voter was significantly more liberal than the typical suburban or rural voters in those regions. Suburban voters were somewhat more liberal than urban voters. Reapportionment, then, shifted the voting weight in the liberal direction within the state legislatures in these regions. The relative liberalness of urban voters in the Northeast and North Central is borne out consistently on other policy questions. Urban voters are consistently more liberal than suburban voters, who, in turn, are typically more liberal than rural voters. Increasing the representation of urban and suburban voters e.g., of Stamford and New Haven shifted the median voter within the state legislatures to the left during the 1960s. Within the Northeast the shift is especially pronounced on questions of health care, school aid, government guaranteed jobs, and, in the Northeast, religion. As with the internal politics of the parties, the urban and rural areas of teh West differed little. The only consistent and signficant differences within this region appear on the issues of school segregation and civil rights. Urban voters in the West were much more liberal than suburban and rural voters on this issue. The Southern electorate showed a somewhat different pattern. On general ideology, the differences were slight, but rural southerners did tend to express a slightly more conservative overall identity. On moral and social issues, the South on the whole paralleled the Northeast and North Central. Though not as liberal as the nation as a whole, urban and suburban areas were 12

much more liberal than rural Southerners in their attitudes on school integration, desegregation, and civil rights. Urban and suburban southerners also expressed more liberal opinions about religion in schools and the truth of the bible than their rural counterparts. However, on domestic economic policies and social welfare, urban and rural Southerners cannot be said to have differed consistently. On some questions, such as health care and economic regulation, rural Southerners expressed more liberal attitudes. On government guaranteed jobs, urban southerners were more liberal. And, suburban southerners often expressed the most conservative views within their region on social welfare and economic policy. 5. Discussion The answers to our two puzzles are, we hope, now evident. The expectation that Democrats would gain everywhere was based on the false premise that the Democrats resided in the cities and the Republicans in small towns and farms. The party splits varied across regions. In the South, rural areas were relatively more Democratic than urban areas; hence, malapportionment gave Democrats more representation than they deserved. In the Northeast and North Central, Republicans received greater weight than their numbers justified. Within regions the partisan differences across locales were massive, but averaged across the country the seem modest. Did reapportionment lead to partisan realignments within the regions? The answer is clearly yes. Erickson s (1971) study of the realtionship between seats and votes in state legislatures shows a distinctly Republican in the North and Midwest, which, in turn, produced a Democratic gain. In the West, the pattern is uneven. His sample excludes the South. We replicated his analysis for the Southern states where there was some party competition before 1964 Florida, North Carolina, Tennessee, and Virginia. In each, there is an enormous Democratic bias before 1966, which vanishes after reapportionment. 5 Reapportionment, as Erickson correctly observed, led to more party competition throughout the country in state legislatures. The consequent policy implications of reapportionment, also, differ between the North and the South and West. On economic and social welfare policy, reapportionment likely tilted the Northeastern and North Central state legislatures to the left. In the West and South, geography 5 Results available upon request. 13

bore little relationship to policy preferences on these issues. Past research has tested whether public expenditures and expenditures on social welfare programs grew throughout the country as a result of reapportionment (see Ansolabehere, Gerber, and Snyder 2002). A nation-wide increase, however, is not what the survey data clearly predict. Instead, we expect increases in social welfare spending and overall spending in the Northeast and North Central, but not in the South and West. Preliminary analyses of data on total government spending suggest that this is indeed what happened. On racial matters civil rights, segregation, and and school integration urban and suburban voters expressed consistently more liberal attitudes than rural voters. Reapportionment in the mid-1960s, we believe, strengthened the political hand of those attempting to implement racial integration policies within the states. Likewise, malapportionment served as an obstacle to integration and the civil rights movement. The areas within states and within the parties that expressed the greatest opposition to expanding civil rights and integrating schools and other facilities had disproportionate voting strength in state legislative elections and in the legislatures themselves. Importantly, Erickson (1973) found substantial effects of reapportionment on passage of civil rights legislation in the states. The long-term consequences of reapportionment are more difficult to divine. Shifting policies and party positions may have subsequently changed people s partisan attachments, leading to further shifts in the positions of the parties and the policies produced by state legisaltures. One dynamic to which reapportionment likely contributed was the leftward movement of the Democratic parties within the states and throughout the nation. The shift is foreshadowed in our survey data. Urban Democrats were badly underrepresented in state legislatures and, thus, within their party. The newly elected legislators from new urban seats in the late 1960s and early 1970s represented a markedly different constituencies than the rural seats they supplanted. These new urban Democratic districts, the NES data reveal, were much more liberal than the rural Democratic seats on civil rights, school aid, government jobs, health care, labor relations, and, in the Northeast and North Central, religion. As the party moved left, it lost many of its rural conservative adherents. As the parties realigned in the 1970s, 1980s and 1990s, the urban centers emerged as the new core of the Democratic electorate. 14

.75 LA Correlation Between Average Democratic Vote and Representation in Lower Chamber.5.2 0 -.2 -.5 WY NJ WA OR PA ME KS MT IA NY CA NV OK NC IN KY MI ID MN SD UT NM WV MO OH IL GA MS TX MD SC AL VA AR CO ND WI FL TN -.75 -.75 -.5 -.2 0.2.5.75 Correlation Between Vote and Representation in Upper Chamber Figure 1 15

Table 1: Standardized Party Identification By Region and Locality, National Election Studies, 1952-1968 North Northeast Central South West Urban 12 13 32-2 Suburban -29-20 31-9 Rural -49-23 31 11 16

Table 2: Attitudes by Region and Type of Locality From National Election Studies, 1952-1968, Democrats Northeast North Central South West Item U S R U S R U S R U S R Ideology 71 57 61 62 43 35-1 4-15 49 57 48 Social welfare 48 30 26 39 7 12-0 -3-6 25 54 39 Health care 53 28 43 31 22 5 15 9 20 11 15 28 School aid I 46 30 4 23 24-6 -5-14 -10 18 17 14 School aid II 34 14-0 45 25-9 49 34 11 45 64 41 Govt and jobs 53 9 6 42 19-3 32 0 22 28 21 24 Regulation 18 17 0 10 15 12 2-10 5 42 46 66 Labor influence 65 22 18 16 17-18 32-12 5-9 14 16 Labor thermometer 72 59 42 60 26 27 17 15 3 59 64 45 Power of fedl govt 73 72 63 95 106 76 45 45 19 116 92 104 School integration I 51 43 12 36 20 12-41 -55-70 59 46 37 School integration II 84 61 66 53 38 22-23 -42-87 45 43 33 Segregation I 27-24 -20-6 -38-55 -32-62 -112 12-29 -16 Segregation II 22 10 28-21 -52-53 -49-57 -97-32 -61-63 Civil rights I 3-43 -40 37-10 -18 4-20 -58 32-4 1 Civil rights II 55 10 34 41 12-0 26-0 -19 8-30 -30 Religious schools 28 41 86 59 1 5 21 32 29-8 -43 4 School prayer 24-14 -19 34 42 62-45 -31-47 23 24 75 Truth of bible 20-21 -18-33 -20-40 -22 19-23 -20 21 9 Sample includes all respondents who identify themselves as strong or weak Democrats. Entries give expected score on item, where all items have been normalized to have a mean of 0 and standard deviation of 100 and de-trended. So, e.g. a score of 50 means 1/2 of one standard deviation above the overall mean on the item. Higher scores denote more liberal position on the item. Bold items are for cases where the F-test of the hypothesis of no-difference between urban, suburban and rural respondents is rejected at the.05 level. 17

Table 3: Attitudes by Region and Type of Locality From National Election Studies, 1952-1968, Republicans Northeast North Central South West Item U S R U S R U S R U S R Ideology -60-43 -55-32 -27-44 -86-75 -66-78 -87-68 Social welfare -16 6-20 -17-58 -71-17 -30-3 -23-26 -43 Health care 10-27 -16-44 -68-47 -64-32 -5-11 -49-67 School aid I 35-26 -32-32 -57-46 -31-21 3-10 -53-36 School aid II -11-38 -50-45 -47-55 1-57 -19-45 -42-48 Govt and jobs 8-16 -23-16 -32-28 -46-28 23-15 -42-27 Regulation -34-48 -33-47 -42-18 -79-27 -5-38 -36-60 Labor influence 10-19 -8-13 6-27 6-29 4-28 -41-46 Labor thermometer -35-47 -38-38 -49-51 2 6 36-74 -71-91 Power of fedl govt -7 1-19 -29-13 -16-50 -26-16 -123-104 -100 School integration I 40 39 14 13 14 7-7 -39-30 54 17 50 School integration II 64 64 72 41 33 39-13 -11-33 48 35 52 Segregation I 11 4 0 4-7 9-66 -85-113 22 38 40 Segregation II 46 42 68-1 -6-13 -80-82 -118 20 11 16 Civil rights I -33-32 -33-14 -9-33 -9 4-8 -60-86 -59 Civil rights II -10-29 -33-33 -16-17 -10-68 -23-88 -124-115 Religious schools 30-35 -32-43 -55-32 -71-13 26-51 -32-60 School prayer -33-32 -43-6 -17-28 -47-23 -40-20 29-2 Truth of bible 36-3 -11 29 26-5 45 25-13 16 36 9 Sample includes all respondents who identify themselves as strong or weak Republicans. Entries give expected score on item, where all items have been normalized to have a mean of 0 and standard deviation of 100 and de-trended. So, e.g. a score of 50 means 1/2 of one standard deviation above the overall mean on the item. Higher scores denote more liberal position on the item. Bold items are for cases where the F-test of the hypothesis of no-difference between urban, suburban and rural respondents is rejected at the.05 level. 18

Table 5: Attitudes by Region and Type of Locality From National Election Studies, 1952-1968, All Respondents Northeast North Central South West Item U S R U S R U S R U S R Party 12-29 -49 13-20 -23 32 21 31-2 -9 11 Ideology 31 6 2 40 17 7-3 -9-13 14 13 15 Social welfare 17 23-9 19-31 -22-1 -7-6 10 1 16 Health care 39-4 -1 5-15 -22-2 -7 16 4-9 -3 School aid I 40-0 -15 8-17 -20-10 -16-6 3-0 -5 School aid II 26-2 -18 12 6-17 35 7 13 30 23 12 Govt and jobs 35-4 -15 19-5 -15 17-10 22 10-10 8 Regulation 2-19 -20-4 -21-6 -16-19 -1 13 17 29 Labor influence 31-1 6 8 4-19 22-20 2-19 -6-4 Labor thermometer 28 2 4 26-3 -6 8-0 1 4 4 10 Power of fedl govt 56 40 23 51 52 38 29 20 17 40 15 33 School integration I 48 37 21 26 18 9-22 -51-54 52 34 38 School integration II 86 65 68 55 40 34-19 -37-72 57 47 48 Segregation I 34 3 9-13 -28-32 -35-55 -97 21-0 10 Segregation II 35 21 40-5 -24-27 -46-55 -90-9 -34-35 Civil rights I 3-29 -27 22-3 -16 20-15 -31-4 -34-19 Civil rights II 47 10 19 21 10 3 26-4 2-9 -48-32 Religious schools 36-5 -1 9-18 -13-3 2 28-15 -38-6 School prayer 22-8 -19 22 26 27-37 -26-44 12 28 38 Truth of bible 16-14 -24-7 2-27 -4 16-27 4 36 6 Entries give expected score on item, where all items have been normalized to have a mean of 0 and standard deviation of 100 and de-trended. So, e.g. a score of 50 means 1/2 of one standard deviation above the overall mean on the item. Higher scores denote more liberal position on the item. Bold items are for cases where the F-test of the hypothesis of no-difference between urban, suburban and rural respondents is rejected at the.05 level. 19

REFERENCES Ansolabehere, Stephen, Alan Gerber, and James M. Snyder, Jr. 2002. Equal Votes, Equal Money: Court-Ordered Redistricting and the Distribution of Public Expenditures in the American States. American Political Science Review December. Bicker, William E. 1971. The Effects of Malapportionment in the States A Mistrial. In Reapportionment in the 1970 s, edited by N.W. Polsby. Berkeley: University of California Press. David, Paul T., and Ralph Eisenberg. 1961. Devaluation of the Urban and Suburban Vote. Charlottesville, VA: University of Virginia Press. Dye, Thomas R. 1965. Malapportionment and Public Policy in the States. Journal of Politics 27: 586-601. Dye, Thomas R. 1966. Politics, Economics, and the Public. Chicago: Rand, McNally. Egan, Patrick, Thad Kousser, and Nate Persily. 2002. The Complicated Impact of One Person One Vote on Political Competition and Representation North Carolina Law Review 80: 1299. Erikson, Robert S. 1971. The Partisan Impact of State Legislative Reapportionment. Midwest Journal of Political Science 15: 57-71. Erikson, Robert S. 1973. Reapportionment and Policy: A Further Look at Some Intervening Variables. Annals of the New York Academy of Science 219: 280-290. Frederickson, H. George, and Yong Hyo Cho. 1970. Legislative Reapportionment and Fiscal Policy in the American States. Western Political Quarterly 27: 5-37. Hanson, Roger A., and Robert E. Crew, Jr. 1973. The Policy Impact of Reapportionment. Law and Society Review 8: 69-93. Hofferbert, Richard I. 1966. The Relation Between Public Policy and some Structural and Environmental Variables in the American States. American Political Science Review 60: 73-82. Jacob, Herbert. 1964. The Consequences of Malapportionment: A Note of Caution. Social Forces 43: 256-61. Lowenstein, Daniel. 1995. Electoral Law. Carolina Academic Press. McCubbins, Mathew D., and Thomas Schwartz. 1988. Congress, the Courts, and Public Policy: Consequences of the One Man, One Vote Rule. American Journal of Political Science 32: 388-415. Robeck, Bruce W. 1972. Legislative Partisanship, Constituency and Malapportionment: The Case of California. American Political Science Review 66: 1246-1255. 20